A couple months ago I launched Bookmarked as a library of book recommendations from interesting people.
Each book recommendation requires collecting a lot of data from different sources. Books all have images, genres, synopses, and people recommending books may share their recommendation on Twitter, on YouTube, on podcasts, or even in the foreword of the book itself.
To streamline this data collection process I’ve been working with Stakwork to build an automated workflow that can collect new book recommendations from interesting people faster than a human can.
This week we did our first test run, and I’m optimistic that it will continue to generate 10-100x cost savings at scale.
This week's workflow cost 1,358 sats (paid over Lightning) and saved me 30 minutes of data collection.
There are a number of steps in the workflow. Some rely on APIs to fetch and filter data, while other parts of the workflow send mini-tasks to be solved by human workers (ex. to determine whether someone’s mention of a book is an actual recommendation or simply a mention).
Pretty cool way to use Bitcoin to cut down on data collection costs and help me scale up the Bookmarked library.